WP 2 Performance Evaluation and Manipulation
Lead PI: Prof. Paolo Guasoni
Classic asset pricing theory ignores the presence of intermediaries such as portfolio managers and institutions, assuming that investors trade optimally according to their preferences. In fact, the bulk of financial assets are managed by intermediaries, who make decisions that are optimal under their management contracts, but not necessarily for their beneficiaries.
Following the seminal work of Jensen (1968) on performance attribution, industry practice has polarised its attention on regression models. Alpha – the intercept of the regression of a manager’s past returns on the benchmark(s) returns – has become a popular synonym of superior performance, and the use of regression models in empirical asset pricing has only reinforced this practice. Yet, several recent studies have challenged the use of alpha, and the related Sharpe ratio, as reliable measures of a managers’ performance. Goetzmann, Ingersoll, Spiegel, and Welch (2007) examine dynamic performance manipulation, and conclude that unskilled managers can game common performance measures, including alpha. Common performance measures are also prone to manipulation through the use of derivatives. Hill, Balasubramanian, Gregory, and Tierens (2006) and Lo (2001) show how portfolios with fixed short positions in call and put options can produce the appearance of outperformance, in the absence of any managerial skill.